home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!ogicse!uwm.edu!wupost!decwrl!concert!glv!usenet
- From: bob@bobsled.glv.com (Bob Lunney)
- Newsgroups: comp.databases
- Subject: Re: non-centralized concurrency control
- Message-ID: <1992Aug19.172833.26455@glv.uucp>
- Date: 19 Aug 92 17:28:33 GMT
- Article-I.D.: glv.1992Aug19.172833.26455
- References: <1992Aug18.192031.18140@lmpsbbs.comm.mot.com>
- Sender: usenet@glv.uucp
- Distribution: usa
- Organization: Encompass
- Lines: 24
- Nntp-Posting-Host: 192.111.49.52
-
- In article <1992Aug18.192031.18140@lmpsbbs.comm.mot.com>
- setzer@ssd.comm.mot.com (Thomas Setzer) writes:
- > Does anyone know where I can get some info (textbook or paper or firsthand
- etc,
- > etc) about non-centralized concurrency control mechinisms and
- implementations.
- >
- > What I mean by this is that there are multiple users accessing a database
- > but not having to go through a lock manager. The locking is handled(I should
- > say needs to be, as it is still being designed) by each individual task or
- > process. Any information on performance, contingency, implimentation or
- theory
- > would be great.
- >
- [lots of stuff about tree based locking deleted...]
-
- I attended a very interesting seminar at the University of North Carolina,
- Chapel Hill (UNC) on this subject. The speaker was David B. Johnson of the
- Department of Computer Science at Rice University. His methodology is
- basically a network model wherein a process requests state information from
- other processes that have a relationship with it. The work was done on Sun
- 3/60s running the V operating system (a research-oriented distributed OS). The
- paradigm, however, appears portable to any reasonable (read preemptive
- multitasking) OS.
-